Damage Identification in thin-walled girders through a finite element-based digital twin

Silionis, Nikolaos Ε.; Σιλιώνης, Νικόλαος Ε.

dc.contributor.author	Silionis, Nikolaos Ε.	en
dc.contributor.author	Σιλιώνης, Νικόλαος Ε.	el
dc.date.accessioned	2021-02-02T09:00:08Z
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/52845
dc.identifier.uri	http://dx.doi.org/10.26240/heal.ntua.20543
dc.rights	Default License
dc.subject	Structural health monitoring	en
dc.subject	Damage identification	en
dc.subject	Artificial neural networks	en
dc.subject	Finite element analysis	en
dc.subject	Genetic algorithms	en
dc.subject	Παρακολούθηση κατασκευαστικής υγείας	el
dc.subject	Διάγνωση βλαβών	el
dc.subject	Τεχνητά νευρωνικά δίκτυα	el
dc.subject	Μέθοδος πεπερασμένων στοιχείων	el
dc.subject	Γενετικοί αλγόριθμοι	el
dc.title	Damage Identification in thin-walled girders through a finite element-based digital twin	en
dc.contributor.department	Τομέας Θαλασσίων Κατασκευών, Εργαστήριο Ναυπηγικής Τεχνολογίας	el
heal.type	bachelorThesis
heal.classification	Structural Health Monitoring	en
heal.classification	Structural Analysis	en
heal.dateAvailable	2022-02-01T22:00:00Z
heal.language	en
heal.access	embargo
heal.recordProvider	ntua	el
heal.publicationDate	2020-11-25
heal.abstract	Humanity stands on the verge of an industrial revolution which will drastically alter the way people live, work, and interact with one another. At the core of this stands the digital twin, a virtual representation of physical assets aiming to enable the condition-based monitoring of all their operational aspects. This thesis draws inspiration from this concept and aims to provide an initial approach towards the implementation of a digital twin for ship hull structures, albeit in a simplified form. The main goal of this work is to develop damage-identification methods for thin-walled girders, based on machine learning and optimization concepts. The choice of a thin-walled girder subjected to three-point bending to model the hull structure under still water loads was based on the principles of hull girder strength. A FE-based Digital Twin of a thin-walled girder subjected to three-point bending was developed and used to determine the features of the Structural Health Monitoring system used to facilitate damage detection. This system’s capabilities were tested by the inclusion of a feature simulating the effects of damage on the strain field, known in this work as a Strain Field Disturber. After the capabilities of the SHM system’s strain monitoring scheme to detect damage were established, two methods aimed at solving the inverse problem of predicting the damaged state of the girder, using only strain data as input were developed to complete the SHM framework. The first utilized techniques found in the field of optimization, specifically Genetic Algorithms, and treated the problem as an optimization problem where damage detection corresponds to the minimization of an appropriate error function. The second used Artificial Neural Networks, trained using data obtained from the digital twin to enable the prediction of damaged states based on strain inputs. The capabilities of both methods were tested within the virtual environment, using the digital twin as the means to provide the requisite data. Finally, the damage identification framework was tested against actual experimental data as well. A series of three-point bending experiments were executed and the strain data obtained from them were used to test the efficacy of the developed methods in real-world conditions.	en
heal.advisorName	Ανυφαντής, Κωνσταντίνος	el
heal.committeeMemberName	Τσούβαλης, Νικόλαος	el
heal.committeeMemberName	Παπαλάμπρου, Γεώργιος	el
heal.academicPublisher	Εθνικό Μετσόβιο Πολυτεχνείο. Σχολή Ναυπηγών Μηχανολόγων Μηχανικών. Τομέας Θαλάσσιων Κατασκευών. Εργαστήριο Ναυπηγικής Τεχνολογίας	el
heal.academicPublisherID	ntua
heal.numberOfPages	135 p.	en
heal.fullTextAvailability	false